Fix specifier matching when the specifier is long and has an epoch

`_pad_version` assumes that all components that aren't numeric are
suffixes. But that assumption breaks when epoch numbers are present,
because `_version_split` outputs a component like "2!1".

Fix the assumption by making `_version_split` separate the epoch number
into its own component. Introduce `_version_join` to correctly join
the output of `_version_split` back into a string, which is needed by
`_compare_compatible`.

Fixes pypa#683

CHANGELOG.rst

@@ -10,6 +10,8 @@ Changelog
 * Introduce ``metadata.Metadata`` (along with ``metadata.ExceptionGroup`` and ``metadata.InvalidMetadata``; :issue:`570`)
 * Introduce the ``validate`` keyword parameter to ``utils.validate_name()`` (:issue:`570`)
 * Introduce ``utils.is_normalized_name()`` (:issue:`570`)
+* Do specifier matching correctly when the specifier contains an epoch number
+  and has more components than the version (:issue:`683`)
 
 23.1 - 2023-04-12
 ~~~~~~~~~~~~~~~~~

src/packaging/specifiers.py

@@ -383,7 +383,7 @@ def _compare_compatible(self, prospective: Version, spec: str) -> bool:
 
         # We want everything but the last item in the version, but we want to
         # ignore suffix segments.
-        prefix = ".".join(
+        prefix = _version_join(
             list(itertools.takewhile(_is_not_suffix, _version_split(spec)))[:-1]
         )
 
@@ -404,13 +404,13 @@ def _compare_equal(self, prospective: Version, spec: str) -> bool:
             )
             # Get the normalized version string ignoring the trailing .*
             normalized_spec = canonicalize_version(spec[:-2], strip_trailing_zero=False)
-            # Split the spec out by dots, and pretend that there is an implicit
-            # dot in between a release segment and a pre-release segment.
+            # Split the spec out by bangs and dots, and pretend that there is
+            # an implicit dot in between a release segment and a pre-release segment.
             split_spec = _version_split(normalized_spec)
 
-            # Split the prospective version out by dots, and pretend that there
-            # is an implicit dot in between a release segment and a pre-release
-            # segment.
+            # Split the prospective version out by bangs and dots, and pretend
+            # that there is an implicit dot in between a release segment and
+            # a pre-release segment.
             split_prospective = _version_split(normalized_prospective)
 
             # 0-pad the prospective version before shortening it to get the correct
@@ -645,7 +645,15 @@ def filter(
 
 def _version_split(version: str) -> List[str]:
     result: List[str] = []
-    for item in version.split("."):
+
+    epoch, sep, rest = version.rpartition("!")
+
+    if sep:
+        result.append(epoch)
+    else:
+        result.append("0")
+
+    for item in rest.split("."):
         match = _prefix_regex.search(item)
         if match:
             result.extend(match.groups())
@@ -654,6 +662,15 @@ def _version_split(version: str) -> List[str]:
     return result
 
 
+def _version_join(components: List[str]) -> str:
+    # This function only works with numeric components.
+    assert all(c.isdigit() for c in components)
+
+    epoch, *rest = components
+
+    return epoch + "!" + ".".join(rest)
+
+
 def _is_not_suffix(segment: str) -> bool:
     return not any(
         segment.startswith(prefix) for prefix in ("dev", "a", "b", "rc", "post")

tests/test_specifiers.py

@@ -369,6 +369,7 @@ def test_comparison_non_specifier(self):
                 ("2!1.0", "==2!1.*"),
                 ("2!1.0", "==2!1.0"),
                 ("2!1.0", "!=1.0"),
+                ("2!1.0.0", "==2!1.0.0.0.*"),
                 ("2!1.0.0", "==2!1.0.*"),
                 ("2!1.0.0", "==2!1.*"),
                 ("1.0", "!=2!1.0"),